Ask a question the customer can answer
Self-reported attribution collects a person’s own account of how they discovered or decided to contact a business. It can reveal referrals, communities, content and offline influences that are missing from a tracked visit.
The wording determines the question you are answering. “How did you first hear about us?” asks about discovery. “What prompted you to contact us today?” asks about the immediate trigger. Neither is identical to the last measured click, so label the answer accordingly.
Customer answer
Where the person remembers hearing about you.
Tracked context
What the available interaction records show.
Interpretation
Explain agreement, difference or uncertainty.
Customer memory and digital tracking describe different evidence. Preserve both rather than forcing one to replace the other.
Choose where the question fits the workflow
A short optional question can fit an inquiry form or a qualification conversation. The right placement depends on how much information the customer is already being asked to provide. Do not make the sales process harder just to fill an attribution column.
For a service business, the question can also be asked naturally during a consultation. Keep the person’s answer distinct from the employee’s interpretation. If someone says “a friend sent me your Instagram,” both the referral and the content may be useful context.
Ask a question that people can answer naturally
“How did you first hear about us?” invites a different answer from “What brought you here today?” Choose the wording that fits the question you want to investigate. Keep a short free-text option so that referrals, offline encounters and unfamiliar sources do not get forced into the nearest campaign label.
A customer may remember a friend’s recommendation while the recorded visit comes from a branded search ad. Those observations can coexist. The recommendation may have created awareness; the search interaction may have helped the person return. Neither observation alone tells you what would have happened without the advertising.
Review repeated answers as qualitative evidence and compare them with the available tracked context. Keep the original response when you normalize categories for reporting. Record when the question was asked, because an answer given during a sales call may differ from one collected at the first inquiry. Do not turn a small set of answers into a precise channel-revenue allocation.
Store enough context to interpret the answer
- The exact question and the customer’s original response.
- A normalized category with an explicit unknown or unclear option.
- The collection point and date, separate from tracked campaign fields.
- A note explaining how this evidence will inform decisions and where it cannot establish causality.
Keep tracked and spoken evidence side by side
| Evidence | What it can tell you | Limit |
|---|---|---|
| Tracked campaign label | The recorded context of a visit or inquiry | May not capture earlier discovery |
| Customer answer | What the person recalls or considers influential | Memory and wording affect the answer |
| CRM history | How the inquiry progressed after capture | Depends on complete stage recording |
| Experiment | Estimated effect under the test design | Depends on validity, power and scope |
Disagreement is a prompt for investigation. A paid-search visit after a referral can be consistent with both records. Forcing the answer into one winning channel destroys information the team could use to understand the buying process.
Code responses without erasing their meaning
- Choose one clear question and record where it is asked.
- Keep the original answer where appropriate for the agreed data handling.
- Apply a small, documented category list for reporting.
- Allow multiple influences or unknown where the answer requires it.
- Review a sample of classifications so different people apply the same rule.
Avoid categories so broad that every answer becomes “online.” Also avoid a complex taxonomy that nobody can maintain. The reporting category should support a real decision, such as whether referrals and a particular content channel frequently appear together.
Use the result as another source of context
Responses can suggest what customers notice, which messages they remember and where to investigate demand. They do not establish how many sales a channel caused. A survey answer also does not make untracked historical revenue fully attributable.
Within Growth Control, this is a measurement practice to scope into the intake or review process where useful. It is not a claim that a built-in attribution survey is automatically enabled in every implementation.
Questions and answers
Should I ask first discovery or last trigger?
Choose the question that serves the decision, or ask them separately where the workflow allows. They describe different parts of the buying process.
What if the answer conflicts with UTMs?
Preserve both sources and their definitions. The person may describe an earlier influence while the UTM describes a later visit.
Can I calculate channel ROI from survey answers?
Not as a causal conclusion by themselves. Survey responses provide context but do not establish the counterfactual outcome without the channel.
Should the question be mandatory?
Consider the information cost in your actual intake. A required attribution question can add friction; the decision should fit the customer experience and purpose.
Put the process into practice
Growth Control implements the tools, workflow and team handoff around an agreed business result.
Explore marketing attribution connected to your sales process